The Next Generation Wireless Networks (NGWNs) are seemed to be heterogeneous networks based on the integration of severalwireless technologies. These networks are required to achieve performances equivalent to classic wireless networks by ensuring thecontinuity of communications and the homogeneity of network management during horizontal and vertical handovers. This task iseven more important when management services, like security and quality of service (QoS), are deployed at access technology level.In this paper, we propose a framework for heterogeneous wireless technology integration based on network architecture skeletonand a handover management mechanism. This framework optimizes the layer-2 handover procedure to achieve performancesrequired by sensitive applications while ensuring the minimization of signaling overhead required for operated networks. As anapplication example, we make use of this framework to propose a heterogeneous network based on WiFi and WiMAX technologies.We present an application example of the framework using the specification of a WiFi-WiMAX network. We propose severalperformance evaluations based on simulation tests based on this application. The latter confirm the efficiency of handover delayoptimization and the minimization of management signaling costs.
1. Introduction
The growth of wireless communication has been, in a fewyears, important thanks to the advantages they offer suchas deployment flexibility and user mobility during com-munications. Several wireless technologies have emerged.These technologies have been designed independently andintended to cover specific service types, user categories,and usability domains. Among these technologies, thereis not one good and generic enough to replace all theothers; each technology has its own merit, advantages, anddevelopment possibilities. For example, 3G technologies, forexample, UMTS and CDMA2000, propose network accessassociated to telephony services. WMAN technologies, forexample, WiMAX and HyperMAN, are used to deployoutdoor metropolitan networks. WLAN technologies, forexample, WiFi, have been developed to be an extension ofalready existing wired LANs; they are also used to deploylocal public wireless networks. In addition, user categories
and usability domains have converged so that terminals andcommunication means have evolved to integrate multipletechnologies.
The result of this evolution is a multitechnology environ-ment that can be exploited to offer an enhanced connectivityto users. The Next Generation Wireless Networks (NGWNs)appear to be the integration of already existing and newlydeveloped wireless technologies that offers a heterogeneousaccess to the same global core network. A multi-technologyterminal will be able to change its access technology eachtime its environment changes. For example, it will beconnected to a WiFi access point when it is in the mall; it willhandover to the WiMAX when it will move to the street andit will use UMTS in the train. This could be a great advancedepending on the adequate mechanisms which are availableto ensure a seamless mobility.
On the other hand, wireless technologies are no longerlimited to be a basic communication medium. They eval-uate by integrating several management services such as
2 Journal of Computer Systems, Networks, and Communications
user authentication, data exchange confidentiality, and QoSmanagement. However, the integration of these services atthe access technology level with specific designs will affect thehandover performances in NGWNs. In fact, the change of theserving Point of Attachment (PoA) requires the renegotiationof management services between the terminal and thenetwork in addition to the redirection of data traffic to thenew terminal location. As a result, the HO execution timemay increase significantly, which should induce significantlatency to exchanged data and even the break of the ongoingsession.
Public wireless networks have to guarantee a good levelof service while insuring the transparency of managementto users. The deployment of such networks using het-erogeneous technologies will require a good connectivityduring handovers, by reducing latency, and the homogeneityof management services such as authentication and QoS.This is possible by deploying anticipation mechanisms thatreduce negotiation exchanges between the terminals and thenetwork, such as context transfer and proactive negotiation[1], and accelerate the redirection data traffic during theexecution of the HO.
Researchers have been interested in this problem and sev-eral papers have proposed models for efficient technology-integration solutions that deal with network access providerrequirements. However, the mobility management offeredby these solutions does not ensure yet seamless handoversduring heterogeneous mobility. Indeed, most solutions offerroaming possibilities based on the sharing of user databases.At best, the integration architectures offer to graft one tech-nology to another and to manage heterogeneous mobilitybased on Mobile IP and extensions. These solutions enablethe optimization of the network reattachment (i.e., the layer-3 HO) by limiting the heterogeneous handover to the re-attachment to the new PoA (i.e., layer-2 HO). This does notsolve the connectivity disruption due to the re-establishmentof network services defined at the technology level. On theother hand, the structure of these technology-integrationsolutions is not suited to heterogeneous mobility. Indeed,the organization of the PoAs in the core network is basedon the access technology they offer rather than the closenessof radio coverage while the executed HOs will be based onthe latter closeness. As a consequence, the HO managementmechanisms based on exchanges between heterogeneousentities will result in a nonnegligible overhead that coulddisrupt the network performances.
In this work, we propose a technology-integration frame-work that provides a new approach to deploy next generationwireless networks. This framework offers a heterogeneousaccess to a global network with optimized mobility perfor-mances regarding HO execution time and signaling cost.The idea is to optimize the layer-2 HO execution in aheterogeneous and homogeneous mobility and to adaptthe network architecture so that this optimization yieldsto a minimum signaling surplus. The framework defines anetwork architecture skeleton and HO management mecha-nisms. They tend to optimize the layer-2 HO execution whileensuring the continuity of management services defined atthe technology-level. In addition, we propose an application
of this framework to an actual wireless network basedon the WiFi and WiMAX technologies. We make use ofthis application to demonstrate the ability of the proposedframework to enable the enhancement of HO performanceswhile ensuring a reduced signaling overhead.
This paper is organized as follows. In Section 2, wepropose an overview of solutions adopted for wireless tech-nology integration. In Section 3, we detail the specificationof the technology-integration framework. We propose, inSection 4, the specification of wireless network based onthe WiFi and WiMAX technologies. We demonstrate theadvantages offered by this architecture based on perfor-mances evaluations in Section 5. We detail how the proposedframework can get along with layer-3 mobility managementmechanisms in Section 6. We propose, in Section 7, adiscussion about heterogeneous technology integration. Wedraw up main conclusions and propose future trends of ourwork in Section 8.
2. Technology Integration in the Literature
Heterogeneous-technology integration has been studied byseveral researches. Most studies focused on networks inte-grating UMTS and data wireless technologies, that is, WiFi[2–6] and WiMAX [7–9]. Two inter working architectureshave been proposed: loosely and tightly coupled architectures[2, 10].
With loosely coupled architecture, the interconnectedtechnologies are considered as independent networks con-cerning the handling of data traffic and the managementof network services such as authentication and QoS. Eachtechnology has a separate user subscription and profilemanagement systems. Roaming privileges are assigned tosubscriptions related to one network. This helps to minimizesession disruption based on the cooperation of account-ing entities. The tightly coupled architecture proposes theintegration of wireless technologies in the same networkarchitecture. This integration may be performed in differentlevels of the management architectures of the consideredtechnologies. User subscriptions and profiles are manage-ment based on common centralized entities. In all cases,user mobility is managed using Mobile IP and its extensions[11].
The main advantage of loosely coupled architecturesis the few modifications to technologies and their corenetwork architectures. However, due to the high level ofintegration, the mobility management mechanisms are notable to optimize significantly the performance of layer-3 handover. Thus, the roaming mechanisms are not ableto reduce sufficiently the session disruption to deal withrequirements of sensitive applications.
The tightly coupled architectures propose integrationat lower level of network architecture. The complexity ofthe implementation increases, and more modifications mustbe operated to technologies and core network architecture.Nevertheless, the lower level of integration ensures a veryinteresting enhancement of HO performances [4, 5]. This isdue to the fact that the inter-working takes place at a point ofthe management architecture closer to the mobile terminal.
Journal of Computer Systems, Networks, and Communications 3
The tightly coupled architecture can significantlyimprove the performance of heterogeneous handovers. Thiscan be even more enhanced by using the ConteXt TransferProtocol (CXTP) [12] in addition to MIP. The CXTPproposes a protocol to transfer mobile terminal contextsbetween Access Routers managing the access control ofa wireless network. CXTP has been designed as a genericprotocol that can accommodate a wide range of services. Thecontext transfer can be reactive, during the HO execution,or proactive from the serving AR to a possible target AR.CXTP can be useful if some network services such as userauthentication and QoS are integrated to the layer-3 level inwireless networks [13]. Consequently, several managementexchanges between a terminal and the Access Router (AR),which controls the access to the network, are required duringthe network entry. Thus, the CXTP enables the reduction ofexchanged messages between mobile terminal and target ARduring the HO execution.
However, the latter optimization limits only the effectsof sub network change during terminal mobility (layer-3HO optimization). Indeed, all the negotiation exchangesand the service establishment procedures defined at access-technology level must be performed during heterogeneoushandover executions.
A solution could be the association of the tightly coupledarchitectures to an optimization of the terminal to technol-ogy association procedure. This optimization will take intoaccount the possible resemblances between the definitionof services and user profiles of technologies to preventthe execution of the negotiations and procedures duringhandover executions. This may be based on managementmechanisms like context transfer or proactive execution ofexchanges.
3. Technology-Integration Framework
This framework aims at defining an optimization of thehandover performances as part of a heterogeneous mobility.
We consider an operator network that offers a reliablenetwork access, to mobile terminals, based on several wirelesstechnologies. Network services, such as user authentication,QoS management, and billing, have to work properly andseamlessly while terminals are moving over the network.We define the network architecture and the position ofmanagement entities that are involved in the handovermanagement procedure.
The proposed framework specifies the skeleton of thenetwork architecture, the definition of mobility contextand the L2-HO management mechanisms. The latter pro-poses the enhancement of L2-HO performances based onmobility-context exchanges.
3.1. Network Architecture Skeleton. The global wireless net-work is organized into access subnetworks, each one gatheringa set of PoAs. We do away with the classic organizationof wireless networks that separates each technology in anautonomous network. PoAs can be gathered in access subnetworks based on the closeness of their wireless coverage
or based on common management requirements. It alsoremains possible to gather PoAs offering the same wirelessaccess technology. We define new management entities: theLayer 2 Access Managers (L2-Acc-Mgrs) that manage terminalmobility over the network. To each access subnetwork isassociated an L2-Acc-Mgr. Figure 1 shows this architecture.
The L2-Acc-Mgr integrates several functions to manageterminal mobility. It acts as a service proxy regardingexchanges between terminals and core network entitiesduring the network entry procedure. For example, terminalauthentication is supported by the L2-Acc-Mgr that actsas AAA-proxy between the terminal and the AAA serverin the core network. At the end of this procedure, theL2-Acc-Mgr maintains the terminal authentication profile(authentication keys) to use it for future purposes.
The L2-Acc-Mgr supports the Neighborhood manage-ment function that maintains the PoAs’ neighborhood. Itprovides a list of PoAs to which a terminal may move whilebeing associated with a particular PoA.
The L2-HO management function integrates the intel-ligence related to the L2-HO management, that is, thetriggering of HO management exchanges, the execution ofexchanges and the management of terminal contexts.
3.2. L2-HO Management Mechanisms. During the networkentry, a terminal associates itself with the network andactivates a set of services and functionalities. The terminalcontext includes the parameters negotiated during the net-work entry and states related to network services used bythe terminal [1]. The acceleration of the establishment ofthis context is required, at the time of handover, to reducethe delay that results from the HO execution phase. Theestablishment of the terminal context on the target PoA,based on already available information, is the solution.
The nature of information elements included in theterminal context defines how it can be exploited to perform acontext re-establishment. This defines values of informationelements to be established, when and how they will beestablished, and the network entities that have to managethese information elements [1]. Authors in [14] propose astudy that define the latter points based on the characteristicsof information elements and particularly:
(i) the scope of the information element,
(ii) the transferability of the information element,
(iii) and the stability of the information element valueover the time.
In the following part, we identify the network entitiesthat will manage the context establishment, the values tobe established, the mechanisms that establish contexts, andfinally when the establishment has to be performed (i.e.,before, during, or after the HO execution), while takinginto account the network architecture decided upon andthe nature of information elements that may be included interminal contexts.
3.2.1. Management of Terminal Contexts. Regarding thescope, a terminal context consists of global session and local
4 Journal of Computer Systems, Networks, and Communications
Layer 2 accessmanagers
Centralizedservers
Core network
Accessrouter
Accessrouter
Accesssubnetwork
PoA
PoA
Layer 2 network entry exchanges
Accesssubnetwork
PoA
Terminal
Figure 1: The L2-Acc-Mgr in the network architecture.
association information elements. The global session infor-mation elements are related to the association establishedbetween the terminal and the core network entities such asAAA servers. The local association information elements arerelated to the association established between the terminaland the serving PoA. When a terminal executes a HO withoutperforming a new network entry, it maintains its globalsession while re-establishing the local association with the newserving PoA.
Then, a context information element is transferableinformation when it remains valid while the terminal changesits serving PoA. Such information element can be reusedwith target PoA to avoid renegotiation during HO execution.Other elements are nontransferable context information, theircurrent value, associated to a serving PoA, cannot beexploited to avoid negotiations between the terminal andtarget PoA to establish a new association. This type of infor-mation has to be re-established through regular exchangesduring the HO execution. Finally, an information elementcan be conditionally transferable if the value associated to theserving AP is not valid for transfer; however, it can be usedto define a new value associated to target PoAs. It is possibleto define translation rules for this specific set of informationelements so as to enable their establishment while avoidingnegotiations during HO execution.
Based on these two classifications we define the contentof terminal contexts and the entities that have to managethese contexts, following the recommendation proposed in[1].
The L2-Acc-Mgr is the most entitled entity to manage thegreater part of the terminal context. First, the global session
information elements are held by the L2-Acc-Mgr thanks tothe service proxy function. Second, local information elementsthat are conditionally transferable may require centralizedinformation related to the neighbor PoAs or the terminal tobe translated for re-establishment. The latter information isheld by the L2-Acc-Mgr, so it is the better able to manageconditionally transferable local information elements. TheHO management function of the L2-Acc-Mgr is responsibleof managing the latter information elements, of the terminalcontext.
The HO management function defines the values forinformation elements to be established by the L2-Acc-Mgr.The latter values will be derived based on the ones usedwith the current association, cached information elements orterminal accounting profile. A Translation function is definedas a part of the HO management function. It is responsibleof defining values to be established for information elementsconstituting the context terminal.
This case can be illustrated over a heterogeneous wirelessnetwork offering access to multi-technology terminals. Amobile terminal can switch between two PoAs offeringheterogeneous technologies. In this case, QoS parameters canbe transferred to re-establish the new association since thetwo wireless technologies do not necessarily use the sameQoS representation. A QoS translation function can solve theconformity problem as most QoS management mechanismshave common bases.
The definition of new values for a context informationelement may result into a synchronization problem betweenthe terminal and the network. Indeed, the terminal mustbe able to integrate the translation subfunction used by the
Journal of Computer Systems, Networks, and Communications 5
L2-Acc-Mgr to define the new information element value.Therefore, the translation rules are defined so that both theterminal and the L2-Acc-Mgr can compute a value thatcorresponds to the new association without performing anyexchange.
The local information elements that have values validfor different local associations (transferable information),are managed by the PoAs. A serving PoA is responsible forredistributing them to target PoAs and caching them.
Finally, there is a set of information elements that currentvalues cannot be exploited to avoid management exchangesbetween a mobile terminal and the network to establish anew association. We name this category: non transferablecontext information. This type of information has to be re-established through regular exchanges during the handoverexecution. We can mention connection parameters used witha terminal, for example, data rate. These parameters dependon the position of the terminal in the cell and the servingAP capacity, and so they have to be negotiated during theassociation.
3.2.2. Context Establishment Exchanges. Two options areavailable for context establishment: the context transfer andthe proactive negotiation [1].
The context transfer is an adequate establishment solu-tion for transferable information elements. It is performedbetween the entity managing the information elementand one or a set of PoAs. In the same way, condition-ally transferable information element re-establishment canbe based on a context transfer mechanism. After beingtranslated, an information element is transferred to targetPoAs.
The context transfer is not the appropriate solutionfor the re-establishment of non-transferable informationelements. An information element might require to be re-established over standard exchanges or the involvementof the terminal in the negotiation or generation process.It remains possible to establish non transferable infor-mation elements using proactive negotiations. The latterare based on the standard exchanges usually performedduring the network entry procedure to generate informationelements.
The adequate time to perform a context establishmentdepends on the stability of the information element valueduring the time. There are static information elementsthat values do not change during the local association anddynamic information elements that values change duringa local association based on network conditions, terminalbehaviors, accounting constraints, and so forth. Proactivecontext establishment can be performed with static infor-mation elements so that it will be available immediatelyat the HO execution. However, proactive establishment isnot excluded with dynamic context. This depends on thefrequency of information element update. If an informationelement is known not to be frequently updated, it remainspossible to perform a conditional proactive establishment.The information element shall be associated to a validitycondition. At the time of the handover, the informationelement is used only if the validity condition is verified. In
other cases, the information element is established reactivelyduring HO execution based on its last update.
3.2.3. HO Establishment Exchanges. Regarding our speci-fication, the context transfer is suitable for informationelements managed by the L2-Acc-Mgr. Proactive and reactiveexchanges are combined to manage static and dynamic infor-mation elements. The exchange (a) of Figure 2 shows theproactive establishment procedure involving the L2-Acc-Mgrand two neighbor PoAs. The target PoA may execute a reac-tive exchange to obtain values related to dynamic informa-tion elements from the L2-Acc-Mgr as shown in Figure 2(b).
The establishment of local association information ele-ments managed by serving PoA can be based on contexttransfer and/or proactive negotiation. These mechanisms maybe combined to establish one or more information elementsin the same procedure or used as alternatives for the sameinformation element to define different procedures sincethey have different properties [1]. Figure 3 shows exchangesbased on the two mechanisms.
The context transfer can be proactive and/or reactive. Forthe proactive one, the establishment exchanges are initiatedby the serving PoA with a list of neighbor PoAs indicated bythe L2-Acc-Mgr. During HO execution, a target PoA mayrequire additional information elements from the servingPoA. As such, it can engage reactive context transfers withthe previous serving PoA.
Proactive negotiations are engaged between the termi-nal and neighbor PoAs through the current association(established with the serving PoA). It is mostly used forinformation elements managed by PoAs that cannot beestablished through context transfer.
The L2-Acc-Mgr is responsible of managing L2-HOmanagement exchanges with entities associated to its accesssubnetwork (i.e., PoAs and terminals) and L2-Acc-Mgrsfrom other access subnetworks. Consequently, the L2-HOmanagement exchanges are limited to the access subnetworkduring intrasubnet mobility. Intersubnetworks exchanges arerelayed by L2-Acc-Mgrs during inter-subnetwork mobility.A target L2-Acc-Mgr converses with the serving L2-Acc-Mgr for centralized establishment exchanges as shown inFigure 4.
In a nonoptimized architecture, the HO managementexchanges between PoAs are routed through the core net-work from one access subnetwork to another during inter-subnet mobility. The HO management exchanges betweenPoAs and centralized entities, during an intra-subnet mobil-ity event, are engaged through the core network whilethe terminal mobility is restricted to the access network.Thus, the use of L2-Acc-Mgrs restricts as much as possiblethe HO management operations to intra-access subnetworkexchanges. This may ensure the efficiency of these exchangesand reduce the signaling overhead over the core network.
4. WiFi-WiMAX Network
As an application of the technology-integration framework,we propose the integration of the WiFi and WiMAX
6 Journal of Computer Systems, Networks, and Communications
L2-Acc-Mgr
(2)
Access subnetwork
(3)
(1)
Serving PoA
Neighbor PoA
Target PoA
Reactive exchange
Proactive exchange
(a) proactive establishment
L2-Acc-Mgr
Access subnetwork
(4)
Serving PoA
Neighbor PoA
Target PoA
Reactive exchange
Proactive exchange
(b) reactive establishment
Figure 2: Centralized establishment.
Access subnetwork
Serving PoA
Neighbor PoA
Proactive negotiation
Context transfer
(a) Proactive negociation
Access subnetwork
Serving PoA
Neighbor PoA
Proactive negotiation
Context transfer
(b) Context Transfer
Figure 3: Distributed establishment.
technologies in a heterogeneous wireless network. Thisnetwork offers to terminals a wireless connectivity adaptedto their location. The WiMAX is deployed for an outdooraccess and the WiFi in building for indoor access. Terminalswill roam from one technology to another according totheir movements while being attached to the same globalnetwork.
4.1. WiFi-WiMAX Integration in the Literature. Someresearches were interested in the collaboration betweenWiFi and WiMAX technologies. Most of these researcheshave proposed to use the WiMAX technologies as backhaulsupport for WiFi hotspot [7, 15, 16]. Therefore, the designednetworks did not fall within the category of 4G networks, andthe two technologies do not cooperate to offer the wireless
Journal of Computer Systems, Networks, and Communications 7
access to mobile users. More recent research studies wereinterested in the inter-working of the WiFi and the WiMAXas access technologies in the same heterogeneous network.However, the majority of these studies were limited to theenhancement of the HO decision mechanism between thetwo technologies and did not discuss the problems related tothe integration and the collaboration of these technologies inthe same network architecture [17–19].
In [20], authors were interested in inter-working ofthe WiFi and the WiMAX technologies. They proposed asolution to ensure a continuity of QoS management throughthe heterogeneous wireless access. The solution proposes amapping between the QoS management parameters of eachtechnology to ensure seamless change of technologies. Tofix the context of their work, authors tried to define aninterconnection architecture for the network. They proposedthe interconnection of separate WiFi and WiMAX accessnetworks through a core network and to manage the layer-3 HO using Mobile IP. However, no additional managementarrangements were proposed (e.g., collaboration betweenQoS accounting, context transfer between BSs and APs) toenable the use of the QoS mapping through the deployedaccess network.
Thus, at the best of our knowledge, there is no seriouswork that offers a design of a heterogeneous networkintegrating the WiFi and the WiMAX technologies.
4.2. Technologies’ Overview. We propose an overview ofthe WiFi [21] and WiMAX technologies [22]. We focusparticularly on the network architecture and the layer-2network service defined by each technology and the mannersin which they interact with mobility management.
4.2.1. WiFi. The WiFi technology is based on the IEEE802.11 standard that defines the PHY and MAC layers forthe wireless medium. This standard has been completed byseveral extensions that define services such as the QoS man-agement and user authentication. The proposed specificationis limited to the management of these services through thewireless part of the network and has not defined operationsthat involve centralized entities.
User authentication is proposed by IEEE 802.11i exten-sion [23] that defines a robust securing mechanism offeringa privacy equivalent to wired network. It proposes a completesecurity framework defining the security architecture, the keyhierarchy, and the cryptographic mechanisms. The 802.11iauthentication is based on an authentication key hierarchyand key generation exchanges. They establish mutual authen-tication between peers and generate cryptographic suite tosecure data exchanges.
The basic IEEE 802.11 standard offered only a best effortservice to an application flow. The QoS management forthe WiFi technology has been defined by the IEEE 802.11eextension [24]. Two operation modes have been defined:
(i) a per-packet QoS management, the prioritized QoS,based on priorities associated to transmission queueswith different channel access priorities,
Table 1: User priority to traffic class mapping.
User Priority Traffic Type Description
1 Background Bulk transfers, games, etc.
2 Spare
0 Best Effort Ordinary LAN priority
3 Excellent Effort Best Effort for important users
4 Controlled Load Some important applications
5 Video Less then 100 millisecond delay
6 Voice less than 10 millisecond delay
7 Network Control High requirements
(ii) a per-flow QoS management, the parameterized QoS,based on QoS parameters associated to virtual trafficstream. The latter are a set of data packets to betransferred in accordance with the QoS requirementsof an application flow.
The WiFi equipments and deployed networks are fol-lowed by particular evolution. Indeed, the QoS managementproposed by IEEE 802.11e was not adopted in networkdeployments. The enhancements of the communicationperformances were based on the evolution of the PHY layerperformances.
With the WiFi-WiMAX integration, the WiFi technologywill coexist with the WiMAX technology, which offers astrong service differentiation between categories of datatraffics based on user profiling (c.f. the next subsection). Soas to offer a homogenous network access service to users overthe network, we propose to adopt a QoS-enabled WiFi accessin our specification. We consider the Parameterized QoS asit most closely matches the QoS management defined byWiMAX [25].
The Parameterized QoS proposes a QoS managementbased on virtual connections: the Traffic Streams (TSs). Thelatter are sets of data packets to be transferred in accordancewith the QoS requirements of an application flow. A terminalspecifies TS requirements to the Access Point (AP) usingthe admission control exchange. The requirements can bedata rate, packet size, service interval, and so forth. An APmay accept or reject new Traffic Specification requests basedon the network conditions, terminal profile, and so forth.The traffic differentiation is based on traffic specification(TSPEC) associated to TSs. The TSPEC element contains aset of QoS parameters that define the characteristics and theQoS expectations of a traffic flow. In addition User Priorities(UP) are used to indicate the traffic class of the TS. Table 1presents the mapping between UP values and traffic class.
The WiFi technology was developed to be an extensionof wired networks and not as an operator technology such asWiMAX or UMTS. Thus, the IEEE 802.11 standard and itsextensions have not specified the core network architecturesand mechanisms. The deployment of RSN security andparameterized QoS requires an AAA server that manages theidentities and the profiles of authorized users.
The negotiations defined by the WiFi authenticationand the parameterized QoS, during the network entry,require considerable time, which turns into a connection
8 Journal of Computer Systems, Networks, and Communications
Centralizedserver
Core networkServingL2-Acc-Mgr
TargetL2-Acc-Mgr
Accesssubnetwork
ServingPoATarget PoA Target PoA
TargetPoA
Accesssubnetwork
Terminal
Inter-subnet exchangesIntra-subnet exchanges
Figure 4: HO management exchanges.
interruption during a handover. The authentication processcan last up to 1 s [26]. Several solutions are available toensure reduced authentication delays during horizontal HOless than 25 milliseconds (ms) [27]. However, these solutionsare not effective for a heterogeneous HO management, whichwill be the current architecture results to a new network entryfor the target technology.
4.2.2. WiMAX. The WiMAX technology offers a last milewireless broadband access as an alternative to cable and DSL.It defines the physical layer design and the wireless mediumaccess mechanism and network services such as the QoSmanagement, mobility management, user authentication,and accounting for wireless part of the network based onthe IEEE 80216 standards [28, 29]. In addition, an end-to-end network specification is proposed by the WiMAX forum[30–33]. It includes the core network architecture referencemodels, protocols for end-to-end aspects, procedures forQoS management, and user authentication.
The reference model defines a logical modeling of thenetwork architecture. The Access Service network (ASN)is defined as a set of network functions providing radioaccess to mobile stations. The Connectivity Service Network(CSN) is a set of network functions that provides IP con-nectivity services to Mobile Stations such as IP parametersallocation, Policy and Admission Control, and Inter-ASN
mobility management. CSN includes network elements suchas routers, AAA proxy/servers, and user databases. The QoSmanagement is defined by the NWG specification [30–33]and the IEEE 802.16e-2005 standard [29]. It defines thedata traffic differentiation mechanism over the wireless linkand associated management functions included in the corenetwork entities, that is, ANS-GWs and Authorization andAccounting servers.
A terminal is associated with a number of serviceflows characterized by QoS parameters. This informationis provisioned in a subscriber management system or in apolicy server, typically a AAA server. A service flow is a MACtransport service that provides unidirectional transport ofpackets (uplink or downlink). IEEE 802.16 specifies five DataDelivery services in order to meet the QoS requirement ofmultimedia applications: Unsolicited Grant service (UGS),Real-Time Polling Service (rtPS), Non-Real Time PollingService (nrtPS), Extended Real-Time Variable Rate (ERT-VR)service, and Best Effort (BE). Each Data Delivery Serviceis associated with a predefined set of QoS-related serviceflow parameters. The QoS profile, which is a set resource-access authorizations and preprovisioned service flows, isdownloaded from the AAA server to the ASN-GW atthe network entry as a part of the authentication andauthorization procedure. Service flows creation is initiatedbased on negotiation exchanges engaged by the terminal, theBS, and the ASN-GW.
Journal of Computer Systems, Networks, and Communications 9
Core network
Accesssubnetwork
BS
AP
Accesssubnetwork
(a) heterogeneous access subnetworks
Core network
WiFisubnetwork
Accessrouter
WiMAXsubnetwork
(b) homogeneous access subnetworks
Figure 5: WiFi-WiMAX network.
Security in WiMAX network is based on Key manage-ment protocol (PKM). The latter defines mutual authen-tication exchanges between the terminal and the networkentities, that is, the BSs and the ANS-GWs. These exchangesresult in the generation of a hierarchical sequence ofauthentication keys. Each key is related to the authenticationof the terminal with a level of the access network: BS, ASN-GW, and AAA server. After the authentication, the terminalnegotiates with the serving BS a cryptographic suite for eachprovisioned service flows.
The WiMAX network entry procedure requires, as withWiFi, several exchanges for the authentication and theestablishment of provisioned service flows. The technologydefines an HO management mechanism based on proactiveand reactive terminal context transfers from the ASN-GWand the serving BS to target BSs while attempting to ensureminimal delay and data loss during the HO procedure. Theterminal context includes authentication parameters, serviceflow parameters (QoS information, cryptographic informa-tion, classification rules, etc.), and PHY capabilities of theterminal. Having these information elements, a target BS willbe able to associate the terminal during the HO procedurewith the minimum of negotiation exchanges. However, suchas the HO management mechanism defined for the WiFi, thisoptimization is restricted to horizontal HOs.
4.3. WiFi-WiMAX Integration
4.3.1. Network Architecture. We propose a flexible deploy-ment schema for the network architecture. The accesssubnetworks may offer a homogeneous deployment thatgathers PoAs offering the same technology: WiMAX subnet-works including Base Stations (BSs) and WiFi subnetworksincluding Access Points (APs). It is also possible to offera heterogeneous deployment that gathers PoAs accordingto the wireless coverage neighborhood apart from their
technologies. In all types of deployment, a mobile terminalmay execute vertical HOs (BS to AP and AP to BS) andhorizontal HOs (AP to AP and BS to BS). Figure 5 shows thetwo deployments.
4.3.2. The L2-Acc-Mgr. L2-Acc-Mgrs, associated to accesssubnetworks, manage the L2-HO for both vertical and hori-zontal HOs. They support WiFi and WiMAX specific functionsthat manage authentication and accounting exchanges withterminals during network entries. An L2-Acc-Mgr acts as anASN-GW for the WiMAX terminals and as an AAA proxyfor the WiFi terminal during the network entries. Thesefunctions allow the L2-Acc-mgr to support layer-2 serviceproxy function.
This specification defines management exchanges bet-ween L2-Acc-Mgr and PoAs (APs and BSs), the intelligencerelated the triggering of exchanges, and the management ofcontext information elements. We limit the description ofthe neighborhood management function to the definition ofRecommended PoA lists. The actual content is to be definedby the network operator that can define the neighborhoodmanagement function based on wireless cell load, networktopology, PoA geographic neighborhood, link status, andmobility behaviors.
The translation functions define the information elementvalues to be established during HO procedures for bothvertical and horizontal HOs. This specification considers theuser authentication, the QoS management and WiMAX PHYlayer enhancement as the services to be managed during theL2-HO preparation procedure. In the next subsection, wedetail the specification of this function.
4.3.3. Terminal Context Translation. For horizontal HOs, thetranslation function provides context information elementsbased on the ones used during actual association. The
10 Journal of Computer Systems, Networks, and Communications
Table 2: QoS mapping between IEEE 802.11e and IEEE 802.16e-2005 classes.
802.16e-2005Data Deliveryservice
802.11e UPs Application
UGS 6,7 Voice
ERT-VR 5 Voice with silencesuppression
RT-VR 4 Video
NRT-VR 3 FTP
BE 1,2,0 Email,Web
computation is based on what is defined by each technologyfor internal HO optimization.
When the context establishment is executed to preparea vertical HO (serving PoA and target PoA with differ-ent technologies), the computation of values of contextinformation elements is less obvious than with horizontalHOs. However, we have found a similitude between theQoS and authentication management of WiMAX and WiFi.Therefore, we define a mapping between the terminal contextof the WiFi and WiMAX that enables the translation functionto define values for WiFi context information-elements(resp., for WiMAX context information-elements) based onvalues related to a WiMAX association (resp., for WiFiassociation).
(a) QoS Information Elements. Regarding QoS management,the traffic differentiation defined by IEEE 802.11e parame-terized QoS mechanism and the WiMAX QoS managementare very similar, particularly Traffic Stream and Service Flowconcepts.
We specify an association between User Priorities used inIEEE 802.11e and IEEE 802.16e-2005 Data Delivery services.These two types of information are used to characterize ineach technology the class of the traffic flow. We suggestthe static association between class of services of bothtechnologies shown in Table 2. Classes are mapped accordingto the key QoS requirement for each Data Delivery Service.As shown in the mapping table, more than one User Prioritycorrespond to UGS and BE data delivery service. Therefore,when the IEEE 802.16e-2005 is the serving technology, wepropose to map Service Flows with data delivery servicecorresponding to UGS into TSs with UP equal to 6 and thosewith data delivery service corresponding to BE into TSs withUP equal to 1.
In addition, we propose a mapping between QoS param-eters associated to each IEEE 802.16e-2005 Data Delivery ser-vice and IEEE 802.11e QoS parameters defined in the TSPECinformation element. The IEEE 802.16e-2005 defines specificQoS parameters for each Data Delivery Service. However,IEEE 802.11e defines a list of parameters used for QoScharacterization that may be more extensive than neededor available for any particular instance of parameterizedtraffic. The specification does not define a correspondencebetween traffic categories (defined using UPs) and possiblelists of associated parameters. To be able to ensure a
mapping between QoS parameters, we propose to considerthe matching defined by the IEEE 802.16e-2005 betweenScheduling services and QoS parameters as a reference inthe translation procedure. The parameters associated to atraffic flow depend on the traffic class associated to it in bothIEEE 802.11 and IEEE 802.16e-2005. We propose a statictranslation procedure between QoS parameters to be used bythe Translation Function. The translation process dependson the QoS information related to the current terminalassociation, that is, the serving technology.
(i) Terminal associated to a IEEE 802.11 PoA: in thiscase, the Parameter Translation Function translatesthe TSPEC list into an SF info list.
Firstly, the UP related to the TS is translated intoa Data Delivery Service in accordance to mappingproposed in Table 2. The retained Data DeliveryService indicates the IEEE 802.11e QoS parametersto be determined using the translation. Secondly,the Parameter Translation Function defines valuesrelated to the Data Delivery Service parameters basedon the mapping in Table 3.
(ii) Terminal associated to IEEE 802.16 PoA: in this case,the Parameter Translation Function translates the SFinfo list into a TSPEC list.
SF info includes the Data Delivery Service andrelated QoS parameters. The Parameter TranslationFunction translates the Data Delivery Service intoa UP based on mapping defined in Table 2. Then,it defines which parameters to be included in theTSPEC and their values.
Table 3 presents the mapping used to compute IEEE802.16e-2005 QoS parameters based on the IEEE 802.11eparameters.
We now discuss some translation choices and differencewith mapping used in the reverse translation (i.e., from802.16e-2005 parameters to 802.11e ones).
(a) Unsolicited Grant Interval parameter indicates thenominal interval between successive grant oppor-tunities for UGS and ERT-VR flows. UnsolicitedPolling Interval parameter indicates the same QoScharacteristic for RT-VR flows. These parameters donot have an equivalent in 802.11e QoS parameters.However, the TSPEC include Maximum ServiceInterval and Minimum Service Interval that defines,respectively, maximum and minimum of the intervalbetween the start of two successive transmissionopportunities. Thus, we use these two parametersto define a mean value corresponding to the IEEE802.16e-2005 parameter: (MinimumServiceInterval+ MaximumServiceInterval)/2. When the currentserving technology is the 802.16e-2005, we mayallocate the same value to Maximum and MinimumService Interval 802.11 parameters. This value talliesto Unsolicited Grant Interval or Unsolicited PollingInterval value depending on Data Delivery Service.
Journal of Computer Systems, Networks, and Communications 11
L2-Acc-MgrMSK
PMK
AK
PMK
AK
ServingBS
TargetAP
TargetBS
AK
MSK
PMK
AK
HO-Req HO-Req(AK)HO-Req(PMK)
MS
Figure 6: Proactive key distribution, Scenario 1.
(b) The correspondence between Traffic Priority andUser Priority is defined only for mapping from 802.11specification to the 802.16 one. In the reverse case, thevalue of the User Priority parameter is obtained basedon the Data Delivery Service as previously indicated.
(c) The Tolerated Jitter parameter do not have anequivalent in 802.11e QoS specification. However, wepropose to compute a corresponding value based onavailable parameters. The jitter value is defined as J =max(D) − min(D) where D is the delay imposed toexchanged data packets. We have D = Dl +Dn, whereDl is local delay due to buffering and scheduling andDn is the network delay due to the transmission of thepacket. We suppose that Dl is negligible compared toDn, and thus the latter equation will beD = Dn. Thus,max(D) corresponds to the Delay Bound 802.11parameter. Additionally, min(D) can be computedbased on the data rate perceived by the 802.11 station.The Parameter Translation Function can obtain aMean Data Rate value based on information gatheredby the L2-Acc-Mgr about mobile connectivity andcell states.
(b) Authentication Information Elements. The authenticationprocedures defined by the WiFi and the WiMAX are bothbased on negotiation exchanges that result to the generationof hierarchical sequences of authentication keys. The twokey sequences are similar and have a common root key, theMaster Session Key (MSK), negotiated between the AAAserver, and the terminal for WiFi and WiMAX. Thus, itis possible to define a mapping between levels of two keysequences.
The WiMAX authentication procedure results to theestablishment of the MSK transferred from the AAA server
to the authenticator. The authenticator computes a PairwiseMaster Key (PMK) and an Authorization Key (AK); ittransfers the AK to the Base Station. A 3-way-handshakeexchange is performed between the terminal and the BSbased on the AK. The exchange results in the generation ofTraffic Encryption Keys (TEK).
The IEEE 802.11i authentication results to an MSKnegotiated between the terminal and the AAA server. Thelatter generates a PMK key, based on the identity of theserving AP, that it transfers to the AP. This key is used toperform the 4-way-handshake between the terminal and theserving AP. This exchange computes the Pairwise TransientKey (PTK) used to secure data transfer.
Conforming to the WiMAX specification, the AK isgenerated by the L2-Acc-Mgr, which acts as an ASN-GW,and delivered to the BS. Similarly, the 802.11 PMK isgenerated by the L2-Acc-Mgr (the 802.11 AAA proxy) anddelivered to the AP. The 802.16 AK and the 802.11 PMKhave the same functionality in authentication procedures. Weconsider these two keys as the starting point to define theinter technology translation for security parameters.
When the terminal is associated with a BS, it sharesan 802.16 PMK with the L2-Acc-Mgr. This key is used tocompute the AK that the L2-Acc-Mgr transfers to the BS.During the HO preparation procedure, the L2-Acc-Mgr usesthe 802.16 PMK to generate keys for target PoAs. 802.16 AKsare generated for BSs, and 802.11 PMK are generated for APs.Figure 6 details related exchanges.
When the terminal is associated with an 802.11 AP, itshares an 802.11 PMK with the L2-Acc-Mg.During the HOpreparation procedure, the L2-Acc-Mgr uses the 802.11 PMKto generate keys for target PoAs. 802.16 AKs are generated forBSs, and 802.11 PMK are generated for APs. Figure 7 detailsrelated exchanges.
12 Journal of Computer Systems, Networks, and Communications
Table 3: QoS mapping between IEEE 802.11e and IEEE 802.16e-2005 classes.
Maximum Sustained Traffic Rate Peak Data Rate The peak information rate in bit per second
Maximum Latency Delay Bound The latency period starting at the arrival of a packet at the MAC tillits successful transmission to the destination
Minimum reserved Traffic rate Minimum Data Rate The minimum data rate required by the traffic flow
Maximum Traffic Burst Burst Size The maximum continuous burst the system should accommodate forthe traffic flow
SDU size Nominal MSDU size Number of bytes in a fixed size packet
Unsolicited Polling Interval (a) The maximum nominal interval between successive polling grantopportunities for the traffic flow
Unsolicited Grant Interval (a) The nominal interval between successive grant opportunities for thetraffic flow
Traffic Priority User Priority (b) The priority among two IEEE 802.16e-2005 service flows identical inall QoS parameters.
Tolerated Jitter (c) The maximum delay variation (jitter) (in milliseconds)
(c) WiMAX PHY Information Elements. The WiMAX tech-nology defines parameters related to PHY-layer capabilitiesof terminal. These parameters have no equivalent in the WiFispecification. Thus, we maintain a caching mechanism forPHY-layer capabilities managed by the translation function.PHY-layer capabilities of terminals are maintained duringthe ongoing session. When preparing an HO with target BSs,if a terminal has never been attached to a BS in previousassociations, the L2-Acc-Mgr sends an HO-Req to targetBSs without these parameters. Additionally, it indicates tothe terminal, in the recommended Candidate PoA List, toexecute proactive exchanges to negotiate these parameterswith target BSs.
4.3.4. Context Establishment Procedure. The L2-HO opti-mization is based on the establishment of terminal contextson target PoAs to avoid their re-negotiation and conse-quently reduce the HO delay. The context establishmentprocedure is mainly proactive. The neighborhood man-agement function provides the Recommended PoA List towhich the establishment is initiated. The QoS parameters,the authentication keys, and the WiMAX PHY profiles areestablished based on a context transfer managed by the L2-Acc-Mgr. The cryptographic suites are established based ona context transfer between the serving PoA and target PoAs(preparation of a horizontal HO) or proactive negotiationbetween the terminal and target PoAs (preparation of avertical HO). The translation function computes values forthe information elements to be established based on theavailable terminal context.
In addition to proactive establishment, the specifica-tion defines reactive establishment exchanges that may beengaged by the target PoA during the HO execution.
Figure 8 shows an example of the proactive phase of thecontext establishment procedure. The terminal is associatedwith a serving AP. The context establishment is performedwith an AP and a BS. When a mobile terminal associates itselfthrough an AP, the context establishment is started using an
HO-Request, which includes QoS information elements sentby the serving AP to the L2-Acc-Mgr. The translation func-tion builds the contexts related to PoAs in the RecommendedPoA List. The HO management function initiates contexttransfer to PoAs using HO Request messages that includesterminal contexts. Based on target PoA responses, whichindicates the support of terminal requirements, the HOmanagement function builds the PoA List that is forwardedto the serving AP. The serving AP transfers the list tothe terminal. The cryptographic suites are established, withavailable PoAs, using a context transfer with target APs and aproactive negotiation with the target BSs.
The previous example describes a preparation procedureperformed with target PoAs in the same access network as theserving PoA. The HO messages are exchanged between PoAs,and the L2-Acc-Mgr managing the subnetwork and contextmessages are exchanged between involved PoAs. When atarget PoA is located in an access network different from theserving PoA one, the HO management exchanges are relayedbetween the serving L2-Acc-Mgr and the target L2-Acc-Mgrto reach the involved entities. The serving L2-Acc-Mgr is themanager of the preparation procedure while the target L2-Acc-Mgr relays the messages between the latter entity and thetarget PoA. Figure 9 shows the exchange.
Regarding context transfers between PoAs and proactivenegotiations between the terminal and the target PoAs, wemake the choice not to execute these exchanges during theinter-subnet preparation procedure. Therefore, the prepa-ration will be limited to centralized exchanges performedbetween the L2-Acc-Mgr and the PoAs. This is justified byresults we have obtained in work related to HO preparationmechanisms proposed for the IEEE 802.11 networks regard-ing velocity support and signaling cost [34]. The evaluationhas shown that exchanges performed between PoAs andparticularly proactive negotiations are not adapted to inter-subnet mobility. In fact, they increase the signaling cost ofthe preparation procedure and reduce the HO performancein high mobility environments.
Journal of Computer Systems, Networks, and Communications 13
L2-Acc-MgrMSK
PMK PMK
AK
ServingAP
TargetAP
TargetBS
PMK
MSK
PMK
HO-Req HO-Req(AK)HO-Req(PMK)
MS
Figure 7: Proactive key distribution, Scenario 2.
L2-Acc-Mgr Target AP Target BSTerminalServing AP
HO-Req(QoS param.)
Translation procedure
Context req (crypt. suite)
HO-Ack
Context Rprt
HO-Req (QoS param., auth. key)
HO-Req (QoS param., auth. key, PHY param.)
HO-Resp
HO-Resp
HO-Ack
HO-AckHO-Resp (PoA List)
(PoA List)
3-way-handshake
Key derivation
Figure 8: Example of context establishment.
4.3.5. HO Execution Optimization. The HO preparationprocedure, presented in previous sections, establishes a set ofcontext information elements and parameters in target PoAs.The exchanges engaged during the HO execution depend onthe information elements that were established proactivelyduring the HO preparation procedure or requested reactivelyduring the HO execution. We present in the following para-graphs possible HO execution scenarios for both WiMAXand WiFi technologies. We consider optimal scenarios wheretarget PoAs were able to acquire all context informationelements.
The establishment of the terminal context results in animportant optimization of the L2-HO execution procedurefor both vertical and horizontal HOs. The terminal nolonger needs to reauthenticate itself and to renegotiate QoSparameters and PHY profile (when the WiMAX is the targettechnology) during the L2-HO execution.
Figure 10 presents a regular WiFi network entry thatmay be executed during a first network association andan optimized reassociation procedure that may be executedduring HO with an AP. In the first case, the terminalperforms a regular 802.11i authentication (2, 3, 4, and 5),
14 Journal of Computer Systems, Networks, and Communications
Figure 10: Association versus Re-association with a WiFi Access Point.
including exchanges with the AAA server, and the 802.11etraffic streams’ establishment (6).
During a HO preparation, a target AP may acquire theTraffic Stream (TS) list and the PMK during the first phaseof the procedure based on exchanges performed with theserving L2-Acc-Mgr. The target AP acquires also the PTKbased on a context transfer or computes this key with aproactive negotiation performed with the AP. Therefore, inthe second case of Figure 10, the terminal starts the HOexecution with the legal IEEE 802.11 re-association andauthentication. Over Authentication Req/Resp, the terminaland the target AP inform each other about the preestablishedkeys. Then, they engage a key-handshake to exchange theGroup Temporal Key (GTK). If this part of the authenti-cation exchange succeeds, the new serving AP sends to theterminal the TS List (including TSPECs), and the latter canstart data exchange.
Figure 11 presents a regular WiMAX network entrythat is executed during a first network association and anoptimized re-association procedure that have to be executedduring an HO with a BS. In the first case, the terminalperforms all steps of regular WiMAX association: synchro-nization (1), ranging (2), basic capabilities negotiation (3),authentication (4,5, and 6), cryptographic key negotiation(7,8), and connection establishment (10,11) [29].
During handover preparation, a target BS may acquireproactively the authentication key AK, the encryption keylist TEK list, the SF list, and the WiMAX PHY capabilitiesof the terminal. So in the second case of Figure 11, TheHO execution starts with a Ranging exchange between theterminal and the target BS. The Ranging Response (RNG-Rsp) indicates the re-entry steps that are omitted thanksto the availability of terminal context information elementsobtained during HO execution. Then, the target BS sends an
Journal of Computer Systems, Networks, and Communications 15
Figure 11: Association versus Re-association with a WiMAX Base station.
unsolicited Registration Response (REG-Rsp) that includesinformation about connections. Finally, the terminal sends aBandwidth Request header with zero BR field to the targetBS that regards this message as a confirmation of successfulre-entry registration.
As shown in Figures 10 and 11, the handover execution issignificantly reduced for both WiFi and WiMAX.
5. Performance Evaluation
In this section, we evaluate the performances of the L2-HOmanagement for WiFi-WiMAX network. This evaluationrequires the definition of parameters and metrics that willconstitute the reference of the evaluation. The evaluationcriteria will highlight both the contributions of new mech-anisms and the limits of their application.
5.1. Handover Delay. The most obvious criterion that mustbe evaluated is the HO delay. The latter is defined as thetime during which the station is not connected to any PoA.Therefore, the HO delay includes the time required to detectthe need to perform a handover, to choose a target PoA, andto perform re-association exchanges.
We adopt the network simulator SimulX [35] thatsupports features that enable the design and the evalu-ation of future communication protocols like cross-layerinteractions, multi-interface inter-working in terminals, andheterogeneous network environments. We have integratedto SimulX the IEEE 802.11 architecture [14] and theWiMAX architecture [36]. Both have been validated throughsimulation tests that result in well-known performances of
Table 4: Handover delay.
Target technology Opt. HO (ms) Non-opt. HO (ms)
WiFi 24, 67 1000
WiMAX 23, 16 700
both technologies. The WiFi-WiMAX architecture and theL2-HO optimization mechanism proposed in this researcheshave been implemented in the simulator based on the latterarchitectures [25].
In the first scenario, we evaluate the HO delay performedwhen we use the L2-HO optimization mechanism. Weconsider a wireless network with a single access subnetworkthat includes all the PoAs (two BSs and two APs). A terminalmoves with a straight path to cross the wireless coverageof all PoAs of the network. We measure the delay involvedby the executed L2-HOs. To show the contribution of L2-HO optimization mechanism, we can compare the inter-technology HO delay to the network entry delay of theWiFi and WiMAX technologies, which correspond to non-optimized HOs.
Table 4 lists HO delay values obtained with differenttypes of HOs. The delay due to non-optimized HOs is eval-uated to 700 ms when the WiMAX is the target technologyand 1000 ms when the WiFi is the target technology. Let’snote that the WiFi handover delay is larger than the WiMAXhandover delay although that nonoptimization handoverexecution of WiMAX seems to engage even more exchangesthan the WiFi handover execution (c.f. Figures 10 and 11).Actually, the detection and the search phases contributelargely to the delay induced to traffic during the handover
16 Journal of Computer Systems, Networks, and Communications
procedure of WiFi. However, these phases are well optimizedin handover procedure of WiMAX. For example, there is nosearch phase at the time of HO as the serving BS sends arecommended neighbor list to terminal. As a consequence,the overall HO delay of WiFi network entry during HO islarger that of the WiMAX.
The L2-HO management mechanisms ensure a uni-form execution time for both intratechnology and inter-technology HOs limited to a mean value of 24,63 ms. Thisis obtained thanks to the context establishment mechanismthat ensures the same optimization of the HO executionregardless of the target PoA type.
In a second phase of this evaluation, we study the effectof wireless cell conditions on the performances of the L2-HO optimization performances. We consider a networktopology integrating six BSs with six APs in each WiMAXcell. The PoAs are attached to two access subnetworks: a WiFisubnetwork and a WiMAX subnetwork relayed through acore network, which hosts also the AAA server. A terminalmoves with a straight path and a velocity of 10 m/s. Wemeasure the HO delay for WiFi to WiMAX and WiMAX toWiMAX handovers.
In WiFi networks, the performance of terminal exchangesdepends on the cell load because of the contention-basedmedium access [27]. In a previous research, we wereinterested in the evaluation of HO performances in WiFinetworks. We showed that the wireless cell load has non-negligible effects on the HO execution performances. Weevaluated a management mechanism that ensures the sameoptimization of HO execution for WiFi terminals. Resultsdemonstrated that such optimization ensures a limitedexecution time (lower than 50 ms) even with high loads.
The performance of WiMAX wireless access is notsensitive to the cell load as the medium access is managed bythe BS that allows transmission opportunities to the mediummodeled by transmission frame [28]. However, two param-eters can have an influence on the performances of HO exe-cution: the IEEE 802.16 frame duration and the contention-based transmission period defined for network entry.
The duration of the IEEE 802.16 frame, which is config-urable, has an effect on the delay between two transmissionopportunities for one terminal, which impacts on the delaysfor exchange between the terminals and the BS. In a previousresearch, we have evaluated the variation of the regularWiMAX network entry as a function of the frame duration.Results have shown that the network entry duration varyfrom 700 ms to 1 s with frame duration that varies from 3 msto 12 ms.
We evaluate the effect of the frame duration of theoptimized WiMAX handover. Figure 12 plots the delay dueto optimized WiMAX handover as a function of the 802.16frame duration. This curve shows that the handover delayincreases when the lEEE 802.16 frame duration increases.However, even with frame duration of 12 ms the handoverdelay remains reasonable and does not exceed the value of50 ms (tolerable threshold of real-time applications).
The second parameter considered for WiMAX cells isthe contention-based transmission period. It is used by aterminal that starts an HO procedure or an association
0 2 4 6 8 10 12 14
802.16 frame duration (milliseconds)
0
5
10
15
20
25
30
35
HO
dela
y(m
ilise
con
ds)
WiMAX-WiMAXWiFi-WiMAX
Figure 12: Effect of the 802.16 frame duration on optimized HOperformances.
0 5 10 15 20
Number of terminals
0
20
40
60
80
100
120H
Ode
lay
(mili
seco
nds
)
WiFi-WiMAXWiMAX-WiFi
WiMAX-WiMAXWiFi-WiFi
Figure 13: Effect of number of terminals on optimized HOperformances.
procedure with a BS. This period has a limited durationduring a single frame. The exchanges over it will be impededby the number of terminals trying to communicate.
To evaluate the effect of the number of terminalsexecuting a network entry on the HO delay, we define asimulation scenario that varies the number of terminalsexecuting HOs in the same contention-based transmissionperiod of a cell, and we measure the average of HO delays.The simulation scenario defines a set of terminal movingat the same velocity, over similar trajectories, and neighborstarting points. The network topology includes six BSs withsix APs in each WiMAX cell.
Figure 13 plots the evolution of the HO delay as afunction of the number of terminals. The curves show anincrease of the HO execution time (WiMAX to WiMAX HOsand WiFi to WiMAX HOs) with the increase of the numberof terminals. This parameter exceeds 50 ms as soon as thenumber of terminals that try to associate exceeds 5.
Journal of Computer Systems, Networks, and Communications 17
5.2. Signaling Cost. We propose to evaluate the signalingoverhead of the HO management mechanism associated tothe WiFi-WiMAX integration network. This evaluation aimsto compare the new architecture with alternative networkdeployments under the same conditions.
We consider a realistic deployment of the WiMAX andWiFi technologies over a city. The WiMAX is used to offer anoutdoor access while the WiFi is used to offer indoor accesses.As shown in Figure 14, the WiMAX access is offered to userover a continuous coverage. The WiFi access is offered viascattered areas over the WiMAX coverage.
We compare the performances of the integration archi-tecture (optimized architecture) to an architecture thatdoes not integrate an L2-Acc-Mgr (non-optimized archi-tecture). In the latter architecture, we suppose that theHO management functions, for example, neighborhoodmanagement and context establishment, are supported bycentralized network servers. In addition, we evaluate theinfluence of the design of access subnetworks (homogeneousdeployment versus heterogeneous deployment) on the HOmanagement signaling cost performances. Four networkarchitectures are considered: non-optimized architecturewith homogeneous deployment, non-optimized architecturewith heterogeneous deployment, optimized architecturewith homogeneous deployment, and optimized architecturewith heterogeneous deployment.
The signaling cost of a management mechanism is thetransmission cost of management messages over the networklinks. We define a signaling cost formula that models thesignaling overhead generated by one HO. This formula takesinto account the proactive exchanges with neighbor PoAsduring the HO preparation and the execution exchanges witha target PoA at the time of HO as shown in (1):
SHO = SHOpreparation + SHOexecution. (1)
We consider three types of network links: the locallinks (between entities in the same access subnetwork), thecore network links, and the wireless links. To each link weassociate a weight that models the cost of transmitting ofone byte over this link. These weights allow to quantifylink transmission costs relatively rather than define absolutevalues. A signaling cost formula is the sum of subformulasthat are products of the messages’ size into the crossed links’weight.
The sub-formula SHOpreparation of (1) (resp., SHOexecution)is different as the HO preparation is engaged from a servingAP or a serving BS (resp., the HO execution is engaged witha target AP or a target BS).
We make use of the VanetMobiSim software to emulatethe terminal mobility over the considered wireless deploy-ment [37]. This software offers the list of executed HOsconsidering a wireless deployment and a mobility model. Thecombination of the signaling cost formulas and the mobilitystatistics allow us to evaluate the signaling cost averageof the HO management over the considered deployment[25]. We assume a mix of three types of mobility model:walking users, slow cars, and fast cars. We consider onehop neighborhood definition. The Recommended PoA list
WiFi AP coverage
WiMAX BS coverage
Figure 14: WiFi-WiMAX wireless coverage.
Het
erog
eneo
us
Hom
ogen
eou
s
Architecture
234
56789
10×104
HO
sign
alin
gco
st
(byt
es∗
wei
ght)
OptimizedNon-optimized
Figure 15: Basic configuration signaling cost.
integrates PoAs whose coverage areas are tangent to theserving PoA one.
In a first evaluation, we consider an arbitrary configu-ration with fixed value for link weight. These values indicatethat the transmission cost of a management message over thecore links is twice the transmission cost over the local links.The transmission cost over the wireless links is fourfold thetransmission cost over local links. With this configuration,Figure 15 plots the measured HO signaling costs related tonetwork architectures.
Both the optimized architecture and the heterogeneousdeployment reduce the signaling cost of an HO. Particularly,a combination of these strategies in the same network offers asignificant reduction of the HO signaling cost. The optimizedarchitecture allows the confining of establishment exchangesat best to an access network and at worst to a connectionbetween two L2-Acc-Mgrs. As a result, there is no moreexchanges with centralized servers for HO management. Onthe other hand, the heterogeneous deployment allows togather neighbor PoAs in the same access network. The useof the latter deployment with a non-optimized architectureenables to reduce inter-PoAs exchanges to the intra-accessnetworks exchanges, which reduces significantly the HOmanagement signaling cost. With an optimized architecture,
18 Journal of Computer Systems, Networks, and Communications
the heterogeneous deployment enables, as well, to confinecentralized exchanges to into one access network.
In a second step, we study the effect of architectureparameters on the HO management signaling cost. We con-sider the core-link weight and the neighborhood definition.
Figure 16 plots the evolution of the handover signalingcost as a function of the core-link weight. Both the optimizedarchitecture and the heterogeneous deployment reduce theeffect of core link cost on the HO signaling cost. The com-bination of an optimized architecture and a heterogeneousdeployment offers the better optimization. These resultsconfirm that the design of a network architecture basedon this combination reduces the consumption of the corenetwork resources by HO management signaling overhead.In fact, the signaling exchanges related to a mobile terminalwill be enclosed in the wireless cells and access subnetworksin its mobility areas. Thus, the proposed designs ensure theenhancement of HO performances while reducing the corenetwork resources.
The enlargement of neighborhood definition is impor-tant to ensure a better mobility support. Indeed, a multiple-hop neighborhood should ensure a good support of fastmoving terminals. However, this neighborhood definitionmay result to an increase of the signaling cost of HOs.To study the effect of the neighbor list size, we assumea second neighborhood definition including PoAs that arereachable within two hops. The neighbors of an AP are theAPs that surround within two hops and the BS that coversthe area if it is reachable by a terminal on two hops. Theneighbors of a BS are the APs on its coverage zone reachableat most with two hops and the BSs in its immediate wirelessneighborhood.
We compare the HO signaling costs of this neighborhooddefinition to those obtained with the one-hop neighborhooddefinition proposed in the basic network configuration. Theresults are shown in Figure 17. Both the optimized archi-tecture and the heterogeneous deployment reduce the effectof the growth of the neighbor-list size on the HO signalingcost. As in the previous evaluation, the combination ofthese network designs offers the better results regarding HOmanagement signaling cost. This combination allows theoperator to design wireless network with better mobilitysupport without increasing the HO management signalingoverhead.
6. Interaction with Layer-3 HandoverManagement Mechanisms
In this study, we are interested in optimization of HOperformances in heterogeneous networks. Our proposalshave been limited to the management of layer-2 handovers(L2-HO). Thus, it seemed interesting to study the interactionof this framework with additional HO management mech-anisms, proposed in the literature, that may be deployedin heterogeneous networks. We consider in particular themobility management based on FMIPv6 and the MediaIndependent Handover (MIH) mechanism proposed by theIEEE 802.21 standard to optimize vertical HOs.
Figure 16: Core Link weight effect on HO signaling cost.
Opt
imiz
edh
eter
ogen
eou
s
Non
-opt
imiz
edh
eter
ogen
eou
s
Opt
imiz
edh
omog
eneo
us
Non
-opt
imiz
edh
omog
eneo
us
Architecture
2
4
6
8
10
12
×104H
Osi
gnal
ing
cost
(byt
es∗
wei
ght)
One hopTwo hop
Figure 17: Neighborhood definition effect on HO signaling cost.
6.1. Collaboration with FMIP. The Fast handover for MobileIPv6 (FMIPv6) [38] proposes an improvement to the MIPv6that reduces the layer-3 handover latency. FMIPv6 definesa collaboration between access routers (ARs) to acceleratethe acquisition of link configuration parameters and the for-warding of data traffic when a terminal executes a handoverfrom a previous AR (PAR) to a new AR (NAR). It enablesthe mobile terminal to learn the IPv6 link configurationparameters (IP subnet) related to links, that it detects, beforeit starts effectively the HO execution. The terminal mayrequest information, about all wireless links, to the currentrouter. The reply can be received on the old link or onthe new link (reactive HO). During the HO execution, theterminal sends a message to the NAR to inform it about themovement.
The framework, proposed in this research, enables twopossible configurations regarding L3-HOs. In the first case,
Journal of Computer Systems, Networks, and Communications 19
access subnetworks offers heterogeneous access technologies,which allow having several technologies on the same IPsubnetwork (with the same prefix). This approach avoidsthe need to define a relation between the L2-HO mecha-nisms and a possible L3-HO, since the latter is no longernecessary. With the other possible configuration, each accesssubnetwork offers a single access technology, that is, WiFiaccess subnetworks and WiMAX access subnetworks. Withthis architecture, a vertical HO leads to a L2-HO associated toan L3-HO. Therefore, in addition to the L2-HO managementmechanism we have defined, there is a need to ensure amanagement of the L3-HO. This can be possible by definingan interaction between the latter mechanism and FMIPv6.The L2-HO management mechanism defines the receptionof neighboring PoAs list with which the HO preparationhas been performed. This list may be used, by the FMIPv6module, to engage the management procedure defined pre-viously with ARs attached to PoAs in the list. Upon receivingan indication of the imminent HO execution, the terminalknows its next AR; so it can prepare the configuration of itsinterface with new IP parameters and wait for the indicationof the L2-HO handover execution success. The latter HOexecution is optimized thanks to the preparation procedureof the L2-HO management mechanism. The link availabilityindication may also be used to trigger the preparation offollowing handovers.
6.2. Collaboration with the MIH. The Media IndependentHandover (MIH), proposed by the IEEE 802.21 [39], definestools to manage multiple interfaces in the same terminal.Particularly, it manages exchange of information elementsbetween the terminal and the network to enhance thedecision and search phases of the handover procedure. Italso helps the preparation of the HO execution betweenheterogeneous technologies. For example, the MIH providesto upper layers, link-layer triggers based on reactive andpredictive local link state changes and network information(load balancing information, operator preferences) thatenhance the HO detection. It also supports the transferof global network information (list of available networks,neighbor maps and higher layer network services) fromnetwork servers to the terminal to help it on the HOpreparation procedure. However, the handover executionoptimization is not part of the MIH functions.
The mechanisms, proposed by the MIH, are complemen-tary to the solution we have proposed. Indeed, it is possibleto make use of the MIH with our solution. Its role will beto manage exchanges between the terminal and the networkentities during the HO preparation procedure and to interactwith heterogeneous interfaces for the optimization of HOexecution based on context information elements establishedproactively.
In the integration example we have proposed in IV, weuse mechanisms offered by WiFi and WiMAX to performactions related to the heterogeneous HO management.The IEEE 802.21 proposes media-dependent interfaces andprimitives to be used with the WiFi and the WiMAXtechnologies. This will make easier the integration of the
MIH to the specification we have proposed. MIH functionscan be used, for example to, transfer the Recommended PoAlist to the terminal during HO preparation.
7. Discussions about HeterogeneousTechnology Integration
It is obvious that the mobility management in the het-erogeneous wireless networks is more complex than classicwireless networks. Indeed, the more we try to optimizethe HO at a low level (to ensure better performances), themore proposed solutions are dependent on the specificitiesof technologies. This makes difficult the optimization ofthe L2-HO between heterogeneous technologies, particularlywhen their designs are based on different principles, forexample, the network accesses (connected mode or sharedaccess mode), core network organization, and so forth. Inthis research, we have been able, as well, to propose alayer-2 handover optimization solution based on generaland technology-agnostic framework. This framework offersmechanisms that optimize the L2-HO delay independently ofthe engaged mobility type (homogeneous or heterogeneous),which is a novel idea.
Another interesting point related to this framework isthe ability of the proposed architecture to facilitate theextension of heterogeneous networks based on additionaltechnologies. In fact, the location of HO managementfunctions at L2-Acc-Mgr allows avoiding the modificationof technology specific network entities, for example, PoAs,and functions, for example, authentication and accountingduring these possible extensions. Modifications are restrictedto the adaptation of the L2-Acc-Mgr and their functions.Let us consider the extension of the WiFi-WiMAX network,we have proposed in Section 4, based on a UMTS access.This will require, first, to define the possible associationsbetween the QoS and security parameters in UMTS, WiFi,and WiMAX to include adequate translation rules at theTranslation function. Second, we have to define at UMTScore network entities that manage terminal active contexts,for example, Radio Network Controllers (RNCs) or ServingGPRS Support Node (SGNC), a context exchange withL2-Acc-Mgrs. Therefore, the latter will be able to executetranslation rules and to engage context establishment overWiMAX BSs and/or WiFi AP.
Based on this framework, it is possible to propose anew organization of heterogeneous networks where hetero-geneous PoAs are gathered in the same access subnetworkbased on the neighbor of their wireless coverage. Although,this organization remains far from current deployments’organization, it is very interesting to consider these aspectsfor future network deployments as we have demonstratedthat such a configuration enables optimized heterogeneousHOs with very low singling overhead, which is not thecase with classic network configuration. At least, networkproviders have to retain that with the growth of heteroge-neous mobility there is a need to consider wireless coverageneighborhood between heterogeneous PoAs to ensure areasonable signaling overhead above the core network.
20 Journal of Computer Systems, Networks, and Communications
Finally, we return to the fact that the use of thisframework remains interesting with classic architectures andthat this configuration does not have as many constraints asis believed. In fact, we can use this framework to propose theinterconnection of local and restricted wireless networks, forexample, a WiFi hotspot or a private WLAN, to a larger net-work such as a WWAN or a WMAN. The L2-Acc-Mgrs willconnect the hotspot to the core network router of the WWANthat manages PoAs with coverage close to the hotspot.
8. Conclusion
In this work, we have been interested in the integration ofheterogeneous wireless technologies in the same network.We have defined a technology-integration framework thatdefines an optimization of both horizontal and vertical HOsbased on context establishment mechanisms in heteroge-neous environments. We have proposed an application ofthis general framework to the deployment of a WiFi-WiMAXnetwork. This application demonstrates the utility of thisframework based on a practical network deployment andenables the performance of evaluation tests. The latter showsan efficient optimization of handover delays associated to aminimization of management signaling costs.
We have shown the interest for network access providersto upside the conventional network architecture by mergingthe backbones of heterogeneous wireless access networks.Thus, PoAs will be gathered based on the closeness of wirelesscoverage, which ensures an efficient optimization of HOperformances with minor signaling overhead. Such networkdeployments are more adapted to Next Generation WirelessNetworks where vertical HOs will be more frequent andtrivialized.
In future work, we are interested in proposing anapplication of this framework for the deployment of com-munication systems for transport context and especially railtransport. The latter are required to operate in extremelyvaried environments, such as urban and suburban environ-ments, countryside, sparsely or very low populated, tunnels,and railway stations. In addition, transport systems have veryhigh constraints regarding transmission delays, robustness,and reliability. On the other hand, the fact that trajectoriesare easily predictable offers interesting perspectives for thecontext management, which raises the interest of adaptingour solution to this particular context.
References
[1] M. Kassab, J.-M. Bonnin, and A. Belghith, “General strategiesfor context re-establishment in IEEE 802.11 networks,” inProceedings of the 8th International Conference on IntelligentTransport System Telecommunications (ITST ’08), pp. 72–77,October 2008.
[2] G. Lampropoulos, N. Passas, L. Merakos, and A. Kaloxy-los, “Handover management architectures in integratedwlan/cellular networks,” IEEE Communications Surveys &Tutorials, vol. 7, no. 4, pp. 30–44, 2005.
[3] C. Makaya and S. Pierre, “An interworking architecture forheterogeneous ip wireless networks,” in Proceedings of the 3rd
International Conference on Wireless and Mobile Communica-tions (ICWMC ’07), p. 16, March 2007.
[4] R. Samarasinghe, V. Friderikos, and A. Aghvami, “Analysisof Intersystem Handover: UMTS FDD & WLAN,” Centre forTelecommunications Research.
[5] S.-L. Tsao and C.-C. Lin, “Design and evaluation of UMTS-WLAN interworking strategies,” in Proceedings of the 56thVehicular Technology Conference, vol. 2, pp. 777–781, Septem-ber 2002.
[6] N. Vulic, I. Niemegeers, and S. H. De Groot, “Architecturaloptions for the WLAN integration at the UMTS radio accesslevel,” in Proceedings of the 59th IEEE Vehicular TechnologyConference (VTC ’04), vol. 5, pp. 3009–3013, May 2004.
[7] Y.-T. Chen, “Achieve user authentication and seamless connec-tivity on wifi and wimax interworked wireless city,” in IFIPInternational Conference on Wireless and Optical Communica-tions Networks (WOCN ’07), pp. 1–5, July 2007.
[8] S. Khan, S. Khan, S. A. Mahmud, and H. Al-Raweshidy,“Supplementary interworking architecture for hybriddata networks (UMTS-WiMAX),” in International Multi-Conference on Computing in the Global Information Technology(ICCGI ’06), August 2006.
[9] Q. Nguyen-Vuong, L. Fiat, and N. Agoulmine, “An architec-ture for umts-wimax interworking,” in Proceedings of the 1stInternational Workshop on Broadband Convergence Networks(BcN ’06), pp. 1–10, April 2006.
[10] G. TS, “3GPP System to WLAN Interworking: System Descrip-tion (Release6),” Tech. Rep., 3GPP TS, March 2004.
[11] C. Perkins, “IP Mobility Support for IPv4,” IETF, August 2002.[12] J. Loughney, M. Nakhjiri, C. Perkins, and R. Koodli, “Rfc
context transfer protocol,” Internet Draft, draft-ietf-seamoby-ctp-11.txt, February 2005.
[13] D. Forsberg, Y. Ohba, B. Patil, H. Tschofenig, and A. Yegin,“Protocol for Carrying Authentication for Network Access(PANA),” Draft IETF (Work in progress), December 2006.
[14] M. Kassab, S. Hachana, J.-M. Bonnin, and A. Belghith, “High-mobility effects on WLAN fast re-authentication efficiency,”in FTDA-DN Workshop, Held in Conjunction with Qshine, July2008.
[15] K. Gakhar, A. Gravey, and A. Leroy, “IROISE: a new QoS archi-tecture for IEEE 802.16 and IEEE 802.11e interworking,” inProceedings of the 2nd International Conference on BroadbandNetworks (BROADNETS ’05), pp. 607–612, October 2005.
[16] D. Niyato and E. Hossain, “Wireless broadband access: WiMaxand beyond—integration of WiMAX and WiFi: optimal pric-ing for bandwidth sharing,” IEEE Communications Magazine,vol. 45, no. 5, pp. 140–146, 2007.
[17] Z. Dai, R. Fracchia, J. Gosteau, P. Pellati, and G. Vivier,“Vertical handover criteria and algorithm in IEEE 802.11 and802.16 hybrid networks,” in IEEE International Conference onCommunications (ICC ’08), pp. 2480–2484, May 2008.
[18] J. Nie, J. Wen, Q. Dong, and Z. Zhou, “A seamless handoffin IEEE 802.16a and IEEE 802.1 in hybrid networks,” inInternational Conference on Communications, Circuits andSystems, pp. 383–387, May 2005.
[19] S.-F. Yang and J.-S. Wu, “Handoff management schemes acrosshybrid WiMAXTM and Wi-FiTM networks,” in IEEE Region 10Conference (TENCON ’07), pp. 1–4, November 2007.
[20] T. Ali-Yahiya, K. Sethom, and G. Pujolle, “Seamless continuityof service across WLAN and WiMAN networks: challengesand performance evaluation,” in Proceedings of the 2ndIEEE/IFIP International Workshop on Broadband ConvergenceNetworks (BcN ’07), pp. 1–12, May 2007.
Journal of Computer Systems, Networks, and Communications 21
[21] “Part II: Wireless LAN Medium Access Control (MAC) andPhysical Layer (PHY) Specifications,” EEE Computer Society,Standard, 1999.
[22] WiMAX Forum, “WiMAX Forum Web page,” September2008, http://www.wimaxforum.org/.
[23] LAN/MAN Standards Committee, “IEEE 802.11i: Amend-ment 6: Medium Access Control (MAC) Security Enhance-ments,” IEEE Computer Society, Standard, April 2004.
[24] LAN/MAN Standards Committee, “IEEE 802.11e Amend-ment 8: Medium Access Control (MAC) Quality of ServiceEnhancements,” IEEE Computer Society, Standard, November2005.
[25] M. Kassab and J.-M. Bonnin, “Optimized layer-2 handoverinWiFi-WiMAX networks,” Research Report, TelecomBretagne, 2009.
[26] M. Kassab, A. Belghith, J.-M. Bonnin, and S. Sassi, “Fastand secure hanfoffs for 802.11 infrastructures networks,”NetCon05 Lannion France, november 2005.
[27] M. Kassab, A. Belghith, J.-M. Bonnin, and S. Sassi, “Fastpre-authentication based on proactive key distribution for802.11 infrastructure networks,” in Proceedings of the 1st ACMInternational Workshop on Wireless Multimedia Networkingand Performance Modeling (WMuNeP ’05), pp. 46–53, October2005.
[28] I. L. S. Committee, “Part 16: Air interface for fixed broadbandwireless access systems,” IEEE Computer Society, Standard,June 2004.
[29] I. L. S. Committee, “Part 16: Air Interface for Fixed andMobile Broadband Wireless Access Systems, Amendment 2:Physical and Medium Access Control Layers for CombinedFixed and Mobile Operation in Licensed Bands,” IEEEComputer Society, Standard, February 2006.
[30] N. WG, “Wimax forum network architecture stage 2:architecture tenets, reference model and reference points, part0,” WiMAX Forum, Wimax End-to-End Network SystemsArchitecture, August 2007.
[31] N. WG, “Wimax forum network architecture stage 3:detailed protocols and procedures,” WiMAX Forum,Wimax End-to-End Network Systems Architecture,March 2007.
[32] N. WG, “Wimax forum network architecture stage 2:architecture tenets, reference model and reference points, part1,” WiMAX Forum, Wimax End-to-End Network SystemsArchitecture, August 2007.
[33] N. WG, “Wimax forum network architecture stage 2:architecture tenets, reference model and reference points, part2,” WiMAX Forum, Wimax End-to-End Network SystemsArchitecture, August 2007.
[34] M. Kassab and J.-M. Bonnin, “HO preparation based onnetwork-entry parameter pre-establishement: a signaling coststudy,” Research Report, Telecom Bretagne, October 2007.
[35] N. Montavont, J. Montavont, and S. Hachana, “WirelessIPv6 simulator: SimulX,” in Proceedings of the 40th AnnualSimulation Symposium, Part of the Spring SimulationMulticonference, Norfolk, Va, USA, March 2007.
[36] M. Kassab, J.-M. Bonnin, and M. Mahdi, “WiMAX Simulationmodule with management architecture and signalingexchanges,” in International Workshop on Network SimulationTools (NSTOOLS), October 2009.
[37] M. Fiore, “Vanetmobisim,” February 2007, http://vanet.eurecom.fr/.
[38] E. R. Koodli, “Mobile IPv6 Fast Handovers,” IETF, RFC 5268,June 2008.
[39] LAN/MAN Standards Committee, “IEEE Standard for Localand Metropolitan Area Networks- Part 21: Media IndependentHandover,” IEEE Std 802.21-2008, pp. c1-301, January 2009.
![21 08 0199-03-0000 Broadcast Handovers Tutorial[1]](https://static.documents.pub/doc/80x56/577d362a1a28ab3a6b925a7b/21-08-0199-03-0000-broadcast-handovers-tutorial1.jpg)
